xml-dev - RE: XML Schemas: the wrong name

[Ed Staub <estaub@kana.com>]

While I agree in principle, I'm not sure whether I do in practice.

It's the 80/20 rule thing again: if deriving API from schema (or vice versa)
does a lot of useful work and eliminates a lot of manual work, why not use
them?  The main thing is that automatic mappings must be overridable by
hand-crafted APIs or schemas when appropriate.  Also, it must be possible to
provide "hints" to automatic mapping mechanisms to make them more broadly
applicable.

If you're involved with Java, I'm curious about your opinion of the JSR-031
("Adelard") effort.

-Ed Staub

-----Original Message-----
From: Steven R. Newcomb [mailto:srn@coolheads.com]
Sent: Wednesday, October 25, 2000 2:44 PM
To: steven.harris@tenzing.com
Cc: xml-dev@xml.org
Subject: Re: XML Schemas: the wrong name


[Steven Harris:]
> In particular, what kind of exchanged information doesn't deserve an
> API? What are the benefits of separating the two?

Egad!  I didn't say that exchanged information doesn't deserve an API.
I'm saying a very, very different thing.

I'm saying that, for many kinds of information, attempting to derive
the API from the syntactic structure, or attempting to define the
syntax in terms of the desired API, are bad ideas.  The robustness of
the syntax will suffer, or the usefulness of the API will suffer, or
both, or you will throw up your hands and get disgusted with this
whole XML shtick, because XML is clearly not strong enough to deal
with your problem in a reasonable way.  Bad outcomes, all, and all
completely unnecessary.

In the general case, for any given kind of information to be
interchanged, we need:

* a model for the syntax of the information, *and*

* a model for the properties of the information, once it has been
  understood.

-Steve

--
Steven R. Newcomb, Consultant
srn@coolheads.com

voice: +1 972 359 8160
fax:   +1 972 359 0270

405 Flagler Court
Allen, Texas 75013-2821 USA

"We're not exactly anti-schema, but we're sure pro-DTD."
 -- doctypes.org